Genetics Of Osteopontin in Patients With Chronic Kidney Disease: The German Chronic Kidney Disease Study

Abstract

Organic Herbs for Kidney Health

Introduction 

Osteopontin (OPN) encoded by the SPP1 gene was first described as a glycoprotein belonging to the SIBLING (Small Integrin-Binding LIgand N-linked Glycoprotein) family in 1985 [1]. OPN is expressed in a multitude of tissues like osteoblasts, osteocytes, odontoblasts (playing a role in mineralization and bone resorption [2,3]) macrophages, smooth muscle cells, and endothelial cells, but can also be found in the inner ear, the central nervous system, and the placenta [1,2]. Although OPN can be detected in many cell types it is predominantly synthesized and expressed in kidney tissue. OPN production is stimulated by many factors including parathyroid hormone, calcitriol, calcium, phosphate, and cytokines. The protein can bind integrins through a specific peptide sequence, the arginine-glycine-aspartic acid (RGD) motif, making interaction with various cell types possible (via the nuclear factor kappa B pathway, [4,5]). In the kidney, integrins can be found in the Bowman's capsule, glomerular epithelium, and vascular epithelium [6,7]. OPN is synthesized in the thick ascending limb of Henle's loop and the distal tubule [1,8].

In a review by Kaleta (2019), known (patho)physiological roles of OPN have been discussed [1]. Based on this review, the physiological role of OPN in the kidney is not fully understood yet, but it has been suggested as being essential for tubulogenesis [1]. SPP1 mRNA as well as OPN protein expression were elevated in mostly rat models of kidney diseases and high OPN expression correlated with proteinuria, reduced kidney function, and fibrosis [1]. One study identified various polymorphisms in the SPP1 promoter region affecting its transcriptional activity [9]. In the past several specific SPP1 gene variants have been associated with the pathogenesis and progression of different kidney diseases. Other case-control studies reported on specific variants in the SPP1 gene being associated with different kidney disease patients in comparison to a (healthy) control group: For example, rs1126616 was repeatedly reported as a marker for lupus nephritis and immunoglobulin A nephropathy [10–14]. In connection with diabetic nephropathy, the two SNPs in SPP1, rs11730582 and rs17524488, have been reported [15,16]. We, therefore, reasoned that the presence of reduced kidney function may represent a good study setting to further establish our understanding of the genetic underpinnings of OPN levels in kidney disease, as some biologic mechanisms might be upregulated and thus be easier to detect, which has been shown before [17–19]. In Jing et al. [19], for example, the magnitude of effects for known loci identified in a GWAS of serum urate in CKD patients were of similar or higher magnitude than those reported from population-based studies. The German Chronic Kidney Disease (GCKD) study comprises a large cohort of CKD patients [20]. Besides demographic and clinical data, genetic data are available as well as baseline measurements of serum OPN, providing an ideal setting to explore the genetics of OPN. For this purpose, we performed a GWAS of serum OPN levels in the GCKD study

Results Description of the GCKD analysis set 

Table 1 gives an overview of baseline characteristics of a selected number of variables for the complete GCKD study cohort and the GWAS analysis set in which participants with complete data on genetics, OPN measurements as well as estimated glomerular filtration rate (eGFR), and urinary albumin-to-creatinine ratio (UACR) are included (S1 Fig). There were no major discrepancies between the complete cohort and the analysis set. Overall, the GWAS analysis set was characterized by a proportion of 60% men with a mean age of 60.2 years (SD: 12.0), with median values of 46.0 mL/min/1.73m2 (p25: 37.0; p75: 57.0) for eGFR and of 50.2 mg/g (p25: 9.4; p75: 382.8) for UACR (Table 1). Among the included participants, 35% had a prevalent diabetes mellitus, 16% were current smokers and 30% reported a history of cardiovascular disease (CVD). Median OPN levels in the complete cohort were 29.2 ng/mL (p25: 20.7; p75: 41.9; Table 1). Levels of OPN increased on average across eGFR categories and UACR categories from a median of 25.4 ng/mL for CKD stage G1/2 to 38.5 ng/mL for CKD stage G4/5, as well as from median OPN values of 25.6 ng/mL for UACR stage A1 to mean OPN values of 34.2 ng/mL for UACR stage A3 (S2 Fig).

Genome-wide association study and fine-mapping 

We conducted a GWAS for serum OPN levels (log2-transformed) using ~7.7 million high-quality autosomal bi-allelic variants of the GCKD study with a minor allele frequency (MAF) of �0.01 (S1 Table). The quantile-quantile plot comparing observed and expected p-values from the OPN GWAS did not indicate inflation (inflation factor λ = 1.01), consistent with the absence of systematic errors (S3 Fig). Overall, the Manhattan plot revealed three genome-wide significant regions associated with OPN levels (p-value <5.0E-08; S4 Fig). Besides the three identified regions, the conditional analysis did not reveal additional independent signals (S5 Fig). The respective association results for the three index SNPs (= SNP with the lowest p-value in the respective region) are presented in Table 2. For all three SNPs, the coded allele was present frequently (allele frequency range 0.5– 0.75). The respective coded allele in our cohort decreased OPN levels on average (S6 Fig) with effect estimates per copy of the coded allele ranging from -0.10 to -0.18 (SE: 0.01–0.02; Table 2). One of the index SNPs on chromosome 4 (rs10011284, 4:88833389) is located upstream of SPP1, which encodes the protein OPN itself (Fig 1A). The other index SNP on chromosome 4 (rs4253311, 4:187174683) maps into the KLKB1 gene (intronic variant), encoding the protein prekallikrein (PK) that is converted to kallikrein (KAL); a protease implicated in the surface-dependent activation of coagulation, bradykinin (BK) release, and potentially the renin-angiotensin-aldosterone system (Fig 2A). The index SNP on chromosome 5 (rs2731673, 5:176839898) maps closest to the F12 gene, which encodes coagulation factor XII, a serine protease that cleaves KLKB1-encoded PK to KAL, among other functions and is also related to blood coagulation, fibrinolysis, and the generation of BK (S7 Fig). A summary of annotations combined from different publicly accessible databases and related to all three SNPs is provided in S2 Table. We next tested whether these three index SNPs were associated with serum OPN levels in the Young Finns Study (YFS) cohort, a population-based study with a mean age of 38 years (SD: 5.0) and a mean eGFR of 92.6 mL/min/1.73m2 (SD: 20.7; S1 Methods). Both SNPs on chromosome 4, rs10011284 and rs4253311, were significantly associated with OPN levels showing also direction consistency (Table 2 and Figs 1B and 2B). In contrast, rs2731673 closest to the F12 gene did not replicate in the YFS cohort (S7 Fig). The two replicated SNPs on chromosome 4 explained 1% of OPN levels each within the GCKD study and did not show a non-additive effect on OPN levels (S8 Fig). Moreover, statistical fine-mapping was performed for the two replicated loci to resolve associated loci into potentially causal variants by constructing credible sets that collectively accounted for a 99% posterior probability of containing the variant or variants that cause the association signal (PPA; Material and methods, [21]). However, fine-mapping results are inconclusive as both constructed sets are large, and single variants included only exhibit low PPA estimates (S3 Table)

Colocalization analyses 

To learn more about the molecular mechanisms and associated phenotypes underlying the identified association signals for OPN, we compared patterns of OPN GWAS results in predefined regions to respective GWAS summary statistics from three other sources using human data (see Material and methods for details). Comparable patterns may indicate a common biological basis. 

Gene expression. First, we performed colocalization analyses of OPN GWAS summary statistics related to the two replicated loci with the corresponding GWAS summary statistics of gene expression in cis using data from the GTEx project and the NEPTUNE study 

(Material and methods). Colocalization (posterior probability of H4 [p12] >0.8) of the OPN association signals were detected with the expression of MEPE in the lung (Fig 3A) and of SPP1 in the pancreas (Fig 3B and S4 Table). Furthermore, colocalization was found between the OPN association signal and expression of F11 in six other tissues (in descending order of H4): tibial artery (Fig 3C), brain cortex, terminal ileum part of the small intestine, muscular of the esophagus, transverse colon, and aortic artery (S4 Table). Except for the colocalization of the OPN signal with an expression of MEPE in the lung, the effect direction of both traits (OPN and gene expression) was concordant (alpha12>0; S4 Table). Plasma proteome. In addition, colocalization analyses were conducted using GWAS summary statistics of SNP associations in cis and in trans with levels of ~3,000 different plasma proteins (pGWAS) reported by Sun et al. (Material and methods, [22]). While no colocalization was present for the summary statistics of the GWAS of OPN and proteins in cis, pGWAS results for 87 proteins from various protein classes were found to trans colocalize with OPN GWAS results for rs4253311 at KLKB1 (S5 Table). For the majority of colocalization results

Fig 3. Comparing summary statistics from OPN GWAS and GTEx tissue colocalizing: (A) MEPE: OPN and lung tissue (H4: p12 = 0.99), (B) SPP1: OPN and pancreas (H4: p12 = 0.85); (C) F11: OPN and tibial artery (H4: P12 = 0.98). Left: scatter plot comparing association p-values from both sources against each other (-log10 scale).Upper right: OPN GWAS results for the region of interest. Lower right: GTEx GWAS results for the region of interest of the respective organ. Colors reflect LD correlation (r2) using 1000G EUR population as reference.

(60/87, 69%), the effect direction of OPN and proteins was concordant, which is by the fact that the activated KKS is involved in a broad spectrum of processes, like inflammation, cancer, cardiovascular disease, as well as (patho)physiological roles in kidney and the central nervous system. For 86 mapped proteins, a Gene Ontology (GO) enrichment analysis was conducted to assess whether their encoding genes were enriched in terms representing specific cellular components and molecular functions (Material and methods, S6 Table). Because of a hierarchical structure of reference lists, implicated terms are partially dependent on each other showing e.g. overlapping upregulated molecular functions as well as upregulated hormone and receptor activities (S9 Fig).

UK Biobank (UKB) diseases.

To address the interplay between genetic modulation of OPN levels and phenotypes we further conducted a colocalization analysis with binary UKB disease traits that showed a genome-wide significant association in the GeneAtlas resource [23]. Based on marginal association statistics, no positive colocalization was detected between OPN and various disease traits (S7 Table). Still, for four of seven traits with an association signal different from the OPN signal in the region (H3: p1.2>0.8), a second independent association signal for the respective disease trait in UKB was detected. We then performed additional colocalization analyses based on the obtained conditional association statistics and identified a positive colocalization between the OPN association signal at rs4253311 at the KLKB1 locus and rs1593 for deep venous thrombosis (DVT; H4: p12 = 0.96; S7 Table and S10 Fig). In the GeneAtlas GWAS of DVT, the major allele of rs1593 (A, allele frequency = 0.87) was associated with a higher risk for DVT (OR = 1.2, p-value 2.4e-16), as was the major allele of rs4253311 (major allele: G, allele frequency = 0.51, OR = 1.13, p-value = 2.6e-16).